During the past decade, the gas cluster ion beam (GCIB) has become well known and widely used for a variety of surface and subsurface processing applications. Because gas cluster ions typically have a large mass, they tend to travel at relatively low velocities (compared to conventional ions) even when accelerated to substantial energies. These low velocities, combined with the inherently weak binding of the clusters, result in unique surface processing capabilities that lead to reduced surface penetration and reduced surface damage compared to conventional ion beams and diffuse plasmas.
Gas cluster ion beams have been employed to smooth, etch, clean, form deposits on, grow films on, or otherwise modify a wide variety of surfaces including for example, metals, semiconductors, and dielectric materials. In applications involving semiconductor and semiconductor-related materials, GCIBs have been employed to clean, smooth, etch, deposit and/or grow films including oxides and others. GCIBs have also been used to introduce doping and lattice-straining atomic species, materials for amorphizing surface layers, and to improve dopant solubility in semiconductor materials. In many cases such GCIB applications have been able to provide results superior to other technologies that employ conventional ions, ion beams, and plasmas. Semiconductor materials include a wide range of materials that may have their electrical properties manipulated by the introduction of dopant materials, and include (without limitation) silicon, germanium, diamond, silicon carbide, and also compound materials comprising group III-IV elements, and group II-VI elements. Because of the ease of forming GCIBs using argon (Ar) as a source gas and because of the inert properties of argon, many applications have been developed for processing the surfaces of implantable medical devices such as coronary stents, orthopedic prostheses, and other implantable medical devices using argon gas GCIBs. In semiconductor applications, a variety of source gases and source gas mixtures have been employed to form GCIBs containing electrical dopants and lattice-straining species, for reactive etching, physical etching, film deposition, film growth, and other useful processes. A variety of practical systems for introducing GCIB processing to a wide range of surface types are known. For example, U.S. Pat. No. 6,676,989 C1 issued to Kirkpatrick et al. teaches a GCIB processing system having a workpiece holder and manipulator suited for processing tubular or cylindrical workpieces such as vascular stents. In another example, U.S. Pat. No. 6,491,800 B2 issued to Kirkpatrick et al, teaches a GCIB processing system having workpiece holders and manipulators for processing other types of non-planar medical devices, including for example, hip joint prostheses. A further example, U.S. Pat. No. 6,486,478 B1 issued to Libby et al. teaches an automated substrate loading/unloading system suitable for processing semiconductor wafers. U.S. Pat. No. 7,115,511 issued to Hautala, teaches the use of a mechanical scanner for scanning a workpiece relative to an un-scanned GCIB. In still another example, U.S. Pat. No. 7,105,199 B2 issued to Blinn et al, teaches the use of GCIB processing to improve the adhesion of drug coatings on medical devices and to modify the elution or release rate of a drug from the medical devices.
Although GCIB processing has been employed successfully for many applications, there are new and existing application needs not fully met by GCIB or other state of the art methods and apparatus. In many situations, while a GCIB can produce dramatic atomic-scale smoothing of an initially somewhat rough surface, the ultimate smoothing that can be achieved is often less than the required smoothness, and in other situations GCIB processing can result in roughening moderately smooth surfaces rather than smoothing them further.
Other needs/opportunities also exist as recognized and resolved through the present invention. In the field of drug-eluting medical implants, GCIB processing has been successful in treating surfaces of drug coatings on medical implants to bind the coating to a substrate or to modify the rate at which drugs are eluted from the coating following implantation into a patient. However, it has been noted that in some cases where GCIB has been used to process drug coatings (which are often very thin and may comprise very expensive drugs), there may occur a weight loss of the drug coating (indicative of drug loss or removal) as a result of the GCIB processing. For the particular cases where such loss occurs (certain drugs and using certain processing parameters) the occurrence is generally undesirable and having a process with the ability to avoid the weight loss, while still obtaining satisfactory control of the drug elution rate, is preferable.
In semiconductor applications, GCIBs have been employed with varying degrees of success in many surface-processing improvements, however opportunities for improvement exist. In conventional GCIB processing, often the result, though significantly improved over earlier conventional technologies, is still not of the quality that is required by the most demanding applications. For example, in smoothing processes, for many materials the final degree of smoothness practically obtainable using GCIB processing does not always meet requirements. In applications where other materials are introduced into semiconductor materials (sometimes called GCIB infusion) for purposes of doping, lattice-straining, and other applications such as film deposition, film growth, and amorphization, the interface between the infused, grown, amorphized, or deposited material often has a roughness or non-uniformity at the interface between the irradiated layer and the underlying substrate that impairs optimal performance of the GCIB-modified layer.
Ions have long been favored for many processes because their electric charge facilitates their manipulation by electrostatic and magnetic fields. This introduces great flexibility in processing. However, in some applications, the charge that is inherent to any ion (including gas cluster ions in a GCIB) may produce undesirable effects in the processed surfaces, GCIB has a distinct advantage over conventional ion beams in that a gas cluster ion with a single or small multiple charge enables the transport and control of a much larger mass-flow (a cluster may consist of hundreds or thousands of molecules) compared to a conventional ion (a single atom, molecule, or molecular fragment.) Particularly in the case of insulating materials, surfaces processed using ions often suffer from charge-induced damage resulting from abrupt discharge of accumulated charges, or production of damaging electrical field-induced stress in the material (again resulting from accumulated charges.) In many such cases, GCIBs have an advantage due to their relatively low charge per mass, but in some instances may not eliminate the target-charging problem. Furthermore, moderate to high current intensity ion beams may suffer from a significant space charge-induced defocusing of the beam that tends to inhibit transporting a well-focused beam over long distances. Again, due to their lower charge per mass relative to conventional ion beams, GCIBs have an advantage, but they do not fully eliminate the space charge transport problem.
A further instance of need or opportunity arises from the fact that although the use of beams of neutral molecules or atoms provides benefit in some surface processing applications and in space charge-free beam transport, it has not generally been easy and economical to produce intense beams of neutral molecules or atoms except for the case of nozzle jets, where the energies are generally on the order of a few milli-electron-volts per atom or molecule, and thus have limited processing capabilities.
In U.S. Pat. No. 4,935,623 of Hughes Electronics Corporation, Knauer has taught a method for forming beams of energetic (1 to 10 eV) charged and/or neutral atoms. Knauer forms a conventional GCIB and directs it at grazing angles against solid surfaces such as silicon plates, which dissociates the cluster ions, resulting in a forward-scattered beam of atoms and conventional ions. This results in an intense but unfocused beam of neutral atoms and ions that may be used for processing, or that following electrostatic separation of the ions may be used for processing as a neutral atom beam. By requiring the scattering of the GCIB off of a solid surface to produce dissociation, a significant problem is introduced by the Knauer techniques. Across a wide range of beam energies, a GCIB produces strong sputtering in surfaces that it strikes. It has been clearly shown (see for example Aoki, T and Matsuo, J, “Molecular dynamics simulations of surface smoothing and sputtering process with glancing-angle gas cluster ion beams,” Nucl. Instr. & Meth. in Phys. Research B 257 (2007), pp. 645-648) that even at grazing angles as employed by Knauer, GCIBs produce considerable sputtering of solids, and thus the forward-scattered neutral beam is contaminated by sputtered ions and neutral atoms and other particles originating in the solid surface used for scattering/dissociation. In a multitude of applications including medical device processing applications and semiconductor processing applications, the presence of such sputtered material contaminating the forward-scattered beam renders it unsuitable for use.
In U.S. Pat. No. 7,060,989, Swenson et al. teach the use of a gas pressure cell having gas pressure higher than the beam generation pressure to modify the gas duster ion energy distribution in a GCIB. The technique lowers the energy of gas cluster ions in a GCIB and modifies some of the surface processing characteristics of such modified GCIBs. Such gas modification of GCIB gas duster ion energy distribution is helpful, but does not reduce problems caused by charges deposited in the workpiece by the ions in the GCIB and does not solve certain processing problems, as for example, the weight loss of drug coatings during GCIB processing. Although the techniques of Swenson et al, can improve the ultimate surface smoothing characteristics of a GCIB, the result is still less than ideal.
Gas clusters and gas cluster ion sizes are typically characterized by N, the number of atoms or molecules (depending on whether the gas is atomic or molecular and including variants such as ions, monomers, dimmers, trimers, ligands) comprising the individual cluster. Many of the advantages contributed by conventional GCIB processing are believed to derive from the low velocities of ions in the GCIB and from the fact that large, loosely bound clusters disintegrate on collision with a solid surface, causing transient heating and pressure but without excessive penetration, implantation, or damage to the substrate beneath the surface. Effects of such large clusters (having N monomers—as defined below—on the order of a few thousand or more) are generally limited to a few tens of Angstroms. However, it has been shown that smaller clusters (having N on the order of a few hundred to about a thousand) produce more damage to an impacted surface and are capable of producing discrete impact craters in a surface (see for example, Houzumi, H., et al. “Scanning tunneling microscopy observation of graphite surfaces irradiated with size-selected Ar cluster ion beams”, Jpn. J. Appl. Phys. V44(8), (2005), p 6252 ff). This crater-forming effect can roughen and remove material from surfaces (etch) in undesirable competition with the surface smoothing effects of the larger clusters, in many other surface processing applications for which GCIB have been found useful, it is believed that the effects of large gas cluster ions and smaller gas cluster ions may compete in counter-productive ways to reduce processing performance. Unfortunately, the readily applied techniques for forming GCIBs all result in generation of beams having a broad distribution of cluster sizes having size, N, ranging from around 100 to as much as several tens of thousands. Often the mean and/or peak of the size distribution lies in the range of from several hundred to a few thousand, with distribution tails gradually diminishing to zero at the size extremes of the distribution. The cluster-ion size distribution and the mean cluster size, NMean, associated with the distribution is dependent on the source gas employed and can be significantly influenced by selection of the parameters of the nozzle used to form the cluster jet, by the pressure drop through the nozzle, and by the nozzle temperature, all according to conventional GCIB formation techniques. Most commercial GCIB processing tools routinely employ magnetic or occasionally electrostatic size separators to remove the smallest ions and clusters (monomers, dimers, trimers, etc. up to around N=10 or more), which are the most damaging. Such filters are often referred to as “monomer filters”, although they typically also remove somewhat larger ions as well as the monomers. Certain electrostatic cluster ion size selectors (as for example the one employed in U.S. Pat. No. 4,935,623, by Knauer) require placing grids of electrical conductors into the beam, which introduces a strong disadvantage due to potential erosion of the grids by the beam, introducing beam contamination while reducing reliability and resulting in the need for additional maintenance to the apparatus. For that reason, monomer and low-mass filters are now typically of the magnetic type (see for examples, U.S. Pat. No. 6,635,883, to Torti et al. and U.S. Pat. No. 6,486,478, to Libby et al.) Aside from the smallest ions (monomers, dimers, etc.), which are effectively removed by magnetic filters, it appears that most GCIBs contain few or no gas cluster ions of sizes below about N=100. It may be that such sizes do not readily form or after forming are not stable. However, clusters in the range from about N=100 to a few hundred seem to be present in the beams of most commercial GCIB processing tools. Values of NMean in the range of from a few hundred to several thousand are commonly encountered when using conventional techniques. Because, for a given acceleration potential the intermediate size clusters travel much faster than the larger clusters, they are more likely to produce craters, rough interfaces, and other undesirable effects, and probably contribute to less than ideal processing when present in a GCIB.
It is therefore an object of this invention to provide apparatus and methods for forming high purity neutral gas cluster beams for workpiece processing.
It is a further object of this invention to provide apparatus and methods to provide high purity gas cluster beams that are substantially free of intermediate size clusters.
Yet another object of this invention is to provide apparatus and methods for forming high purity, focused, intense beams of neutral atoms or molecules with energies in the range of from about 1 eV to as much as a few thousand eV.
Still another object of this invention is to provide apparatus and methods for forming beams capable of improved surface smoothing compared to conventional GCIBs.
An object of this invention is to provide apparatus and methods for forming doped and/or strained films and/or for introducing foreign atomic species into the surfaces of semiconductor or other materials, wherein the processed surface have interfaces to the underlying substrate material that are superior to those formed using conventional GCIB processing.
Another object of this invention is to provide apparatus and methods for forming amorphous regions at the surface of a semiconductor or other material using a Neutral Beam and wherein the interface to the underlying substrate material is superior to one formed using conventional GCIB processing.
A further object of this invention is to provide apparatus and methods for etching surfaces with superior final smoothness as compared to conventional GCIB processing.
Yet another object of this invention is to provide apparatus and methods for forming and/or growing films on surfaces of semiconductor and/or other materials, having interfaces to the underlying substrate material that are superior to those formed using conventional GCIB processing.
Another object of this invention is to provide apparatus and methods for treating drug coatings on drug eluting medical devices to control drug elution rate characteristics with reduced drug weight loss compared to processing with conventional GCIBs.
An additional object of this invention is to provide apparatus and methods for treating electrically insulating materials with Neutral Beams of gas clusters and/or monomers for processing such materials without damage induced by beam transported electrical charges.
One more object of this invention is to provide apparatus and methods for producing very shallow doped layers in semiconductor substrates by implanting high purity neutral monomer beams comprising dopant atoms, wherein the doped layers have much smoother interfaces to the underlying semiconductor material compared to interfaces produced by conventional GICB doping processes.